Search papers, labs, and topics across Lattice.
Qatar University
1
0
2
3
Training VLMs on a unified, multilingual, multitask meme dataset reveals that robust meme understanding requires multimodal training and is highly sensitive to dataset-specific overfitting.
